constraints.md (wo constraint): New constraint for ISA 3.0 (power9).
[gcc]
2016-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
* config/rs6000/constraints.md (wo constraint): New constraint for
ISA 3.0 (power9).
* config/rs6000/rs6000.c (rs6000_debug_reg_global): Add support
for wo constraint.
(rs6000_init_hard_regno_mode_ok): Likewise.
* config/rs6000/rs6000.h (r6000_reg_class_enum): Add support for
wo constraint.
* config/rs6000/altivec.md (altivec_vperm_<mode>): Clean up vperm
expanders not to have constraints. Add support for ISA 3.0 xxperm
instruction. Add support for fusing xxlor with xxperm.
(altivec_vperm_<mode>_internal): Likewise.
(altivec_vperm_v8hiv16qi): Likewise.
(altivec_vperm_<mode>v16q): Likewise.
(altivec_vperm_<mode>_uns): Likewise.
(vperm_v8hiv4si): Likewise.
(vperm_v16qiv8hi): Likewise.
* doc/md.texi (RS/6000 constraints): Document wo constraint.
[gcc/testsuite]
2016-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
* gcc.target/powerpc/p9-permute.c: New test for xxperm code
generation.
Eric Botcazou [Mon, 4 Jan 2016 08:06:52 +0000 (08:06 +0000)]
re PR target/69072 (ICE in function_arg_record_value on 7th packed structure)
PR target/69072
* config/sparc/sparc.c (scan_record_type): Take into account subfields
to compute the PACKED_P predicate.
(function_arg_record_value): Minor tweaks.
gcc/
* doc/extend.texi (Common Function Attributes): Move docs for
MSP430-specific attributes to....
(MSP430 Function Attributes): ...here. Delete the redundant
entries and copy-edit the remaining text.
(MSP430 Variable Attributes): Use uniform format for index
entries and add a cross-reference to the corresponding function
attribute docs.
Jakub Jelinek [Fri, 1 Jan 2016 11:55:59 +0000 (12:55 +0100)]
re PR tree-optimization/69070 (ICE: tree check: expected real_cst, have ssa_name in gimple_expand_builtin_pow, at tree-ssa-math-opts.c:1541 with -fsignaling-nans and powl())
PR tree-optimization/69070
* tree-ssa-math-opts.c (gimple_expand_builtin_pow): Only test
REAL_VALUE_ISSIGNALING_NAN on arg0 if arg0 is a REAL_CST.
Jakub Jelinek [Thu, 31 Dec 2015 23:51:50 +0000 (00:51 +0100)]
re PR target/69015 (ICE: RTL check: expected code 'code_label', have 'return' in find_cond_trap, at ifcvt.c:3715 with -fno-if-conversion and __builtin_trap())
PR target/69015
* ifcvt.c (find_cond_trap): Give up if returnjump_p (jump).
Jerry DeLisle [Wed, 30 Dec 2015 22:10:13 +0000 (22:10 +0000)]
re PR libfortran/68987 (double free or corruption in _gfortran_st_write_done when a write statement to an internal file uses an invalid format and the ERR= specifier appears.)
2015-12-30 Jerry DeLisle <jvdelisle@gcc.gnu.org>
PR fortran/68987
* gfortran.dg/error_format_2.f90: New test.
Michael Meissner [Tue, 29 Dec 2015 17:15:14 +0000 (17:15 +0000)]
rs6000.c (init_float128_ieee): Remove IEEE 128-bit comparison functions in cmp_optab and ucmp_optab.
2015-12-29 Michael Meissner <meissner@linux.vnet.ibm.com>
* config/rs6000/rs6000.c (init_float128_ieee): Remove IEEE 128-bit
comparison functions in cmp_optab and ucmp_optab.
(rs6000_generate_compare): Rewrite IEEE 128-bit floating point
software emulation comparisons to only use __eqkf2, __gekf2,
__lekf2, and __unordkf2 functions.
(rs6000_invalid_binary_op): Add support for -mfloat128-convert.
* config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
__FLOAT128_HARDWARE__ if hardware IEEE 128-bit support is
available.
* config/rs6000/rs6000.opt (-mfloat128-convert): Add debug switch
to allow IBM extended double and IEEE 128-bit floating point to be
converted with default conversions.
* config/rs6000/rs6000.md (extendkftf2): Add converters between
KFmode and TFmode if -mabi=ieeelongdouble.
(trunctfkf2): Likewise.
(ieee128_mfvsrd): Split 64-bit integer conversions into 32-bit and
64-bit insns.
(ieee128_mfvsrd_64bit): Likewise.
(ieee128_mfvsrd_32bit): Likewise.
(ieee128_mtvsrd): Likewise.
(ieee128_mtvsrd_64bit): Likewise.
(ieee128_mtvsrd_32bit): Likewise.
* doc/extend.texi (Floating Types): Document that complex
__float128 does not work currently.
* doc/invoke.texi (RS/6000 and PowerPC Options): Document that
-mfloat128 is only supported on PowerPC 64-bit Linux systems.
re PR fortran/69011 ([OOP] ICE in gfc_advance_chain for ALLOCATE with SOURCE)
gcc/testsuite/ChangeLog:
2015-12-29 Andre Vehreschild <vehre@gcc.gnu.org>
PR fortran/69011
* gfortran.dg/allocate_with_source_16.f90: New test.
gcc/fortran/ChangeLog:
2015-12-29 Andre Vehreschild <vehre@gcc.gnu.org>
PR fortran/69011
* trans-stmt.c (gfc_trans_allocate): Unwrap a NOP_EXPR to make sure
the actual type of the source=-expr is used when it is of class type.
Furthermore prevent an ICE.
Jerry DeLisle [Tue, 29 Dec 2015 05:17:42 +0000 (05:17 +0000)]
re PR libfortran/68987 (double free or corruption in _gfortran_st_write_done when a write statement to an internal file uses an invalid format and the ERR= specifier appears.)
Patrick Palka [Mon, 28 Dec 2015 22:00:14 +0000 (22:00 +0000)]
Tweak the documentation of libiberty's xcrc32 function
libiberty/ChangeLog;
* crc32.c: In the documentation, don't refer to GDB's
now-nonexistent crc32 implementation. In the table-generation
program embedded within the documentation, change the type of
the induction variables i and j from int to unsigned int, to
avoid undefined behavior.
Nathan Sidwell [Mon, 28 Dec 2015 13:38:10 +0000 (13:38 +0000)]
nvptx.c (nvptx_output_call_insn): Expect hard regs.
* config/nvptx/nvptx.c (nvptx_output_call_insn): Expect hard regs.
* config/nvptx/nvptx.md (nvptx_reg_or_mem_operand): Rename to ...
(nvptx_nonimmediate_operand): ... here. Update all uses.
(call_insn_operand): Use REG_P.
(call_operation): Allow hard regs.
* config/rs6000/freebsd64.h: Delete FREEBSD_DYNAMIC_LINKER32/64
defines. Use FBSD_DYNAMIC_LINKER instead.
Rename and simplify LINK_OS_FREEBSD_SPEC_DEF32/64 to
LINK_OS_FREEBSD_SPEC_DEF.
Aditya Kumar [Thu, 24 Dec 2015 16:41:46 +0000 (16:41 +0000)]
Fix computation of single entry/exit of a region.
For basic block with two preds, allow (as single entry) only when the other
edge is a backedge. Similarly for basic block with two succs,
allow (as single exit) only when the other edge is a back edge.
Dominik Vogt [Wed, 23 Dec 2015 11:06:20 +0000 (11:06 +0000)]
Fix broken handling of LABEL_REF in genrecog + genpreds.
gcc/ChangeLog
* config/s390/predicates.md ("larl_operand"): Remove now superfluous
const_int and const_double.
* genrecog.c (safe_predicate_mode): Return false for VOIDmode
LABEL_REFs even if the predicate does not handle const_int,
const_double or const_wide_int.
* genpreds.c (add_mode_tests): Treat LABEL_REF like CONST_INT.
David Sherwood [Wed, 23 Dec 2015 10:28:18 +0000 (10:28 +0000)]
iterators.md (VMAXMINFNM): New int iterator.
2015-12-23 David Sherwood <david.sherwood@arm.com>
gcc/
* config/arm/iterators.md (VMAXMINFNM): New int iterator.
(fmaxmin): New int attribute.
(fmaxmin_op): Likewise.
* config/arm/unspecs.md (UNSPEC_VMAXNM): New unspec.
(UNSPEC_VMINNM): Likewise.
* config/arm/neon.md (fmaxmin): New pattern.
* config/arm/vfp.md (fmaxmin): Likewise.
gcc/testsuite
* gcc.target/arm/fmaxmin.x: New file used by tests below.
* gcc.target/arm/fmaxmin.c: New test.
* gcc.target/arm/vect-fmaxmin.c: Likewise.
H.J. Lu [Wed, 23 Dec 2015 09:49:28 +0000 (09:49 +0000)]
[PATCH] Allow indirect call via GOT for 64-bit Pmode x32
From: H.J. Lu <hongjiu.lu@intel.com>
Since Pmode is 64-bit with -maddress-mode=long for x32, indirect call
via GOT slot doesn't need zero_extend. This patch enables indirect call
via GOT for x32 with 64-bit Pmode.
David Malcolm [Tue, 22 Dec 2015 22:27:45 +0000 (22:27 +0000)]
PR c/68473: sanitize source range-printing within certain macro expansions
gcc/ChangeLog:
PR c/68473
* diagnostic-show-locus.c (layout::layout): Make loc_range const.
Sanitize the layout_range against ranges that finish before they
start.
gcc/testsuite/ChangeLog:
PR c/68473
* gcc.dg/plugin/diagnostic-test-expressions-1.c (fminl): New decl.
(TEST_EQ): New macro.
(test_macro): New function.
* gcc.target/i386/pr68473-1.c: New test case.
David Malcolm [Tue, 22 Dec 2015 22:06:00 +0000 (22:06 +0000)]
libcpp: Avoid unnecessary ad-hoc uses for large source files
libcpp/ChangeLog:
* line-map.c (get_combined_adhoc_loc): Remove condition
on locus < RESERVED_LOCATION_COUNT when considering
whether a caret == start == finish location can be
simply stored as the caret location.
Jakub Jelinek [Tue, 22 Dec 2015 20:47:58 +0000 (21:47 +0100)]
re PR c++/67376 (Comparison with pointer to past-the-end of array fails inside constant expression)
PR c++/67376
* fold-const.c (size_low_cst): Removed.
(fold_comparison): For POINTER_PLUS_EXPR where base is ADDR_EXPR
call get_inner_reference and handle INDIRECT_REF base of it. Use
offset_int for computation of the bitpos.
(fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Formatting
fixes for X +- Y CMP X and C - X CMP X folding. Add X CMP X +- Y
and X CMP C - X folding.
The must-not-throw wrapper for protect_cleanup_actions gets in
the way of being able to access __builtin_eh_pointer without
confusion as the identit of the exception to which we are
referring (b_eh_p has no usable argument up to this point).
Since EH_ELSE never comes from user derived code, let's drop
the c++ specific wrapping.
* tree-eh.c (honor_protect_cleanup_actions): Do not wrap
eh_else in a must-not-throw; set ehp_region for it too.
H.J. Lu [Tue, 22 Dec 2015 15:38:25 +0000 (15:38 +0000)]
[PATCH] Use call-clobbered register for sibcall via GOT
From: H.J. Lu <hongjiu.lu@intel.com>
Since sibcall never returns, we can only use call-clobbered register as
GOT base. Otherwise, callee-saved register used as GOT base won't be
properly restored. sibcall_memory_operand is changed to allow 32-bit
GOT slot only with pseudo register as GOT base for RTL expansion. 2
new patterns, *sibcall_GOT_32 and *sibcall_value_GOT_32, are added to
expose GOT base register to register allocator so that call-clobbered
register will be used for GOT base.
gcc/
PR target/68937
* config/i386/i386.c (ix86_function_ok_for_sibcall): Count
call to global function via GOT slot as indirect call.
* config/i386/i386.md (*sibcall_GOT_32): New pattern.
(*sibcall_value_GOT_32): Likewise.
* config/i386/predicates.md (sibcall_memory_operand): Rewrite.
Allow 32-bit GOT slot only with pseudo register as GOT base.
(GOT32_symbol_operand): New predicate.
Sujoy Saraswati [Tue, 22 Dec 2015 14:04:30 +0000 (14:04 +0000)]
This series of patches fix PR61441.
This series of patches fix PR61441. This patch avoids various transformations
with signaling NaN operands when flag_signaling_nans is on, to avoid folding
which would lose exceptions.
Bootstrapped & regression-tested on x86_64-linux-gnu.
gcc/
* fold-const.c (const_binop): Convert sNaN to qNaN when
flag_signaling_nans is off.
(const_unop): Avoid the operation, other than NEGATE and
ABS, if flag_signaling_nans is on and the operand is an sNaN.
(fold_convert_const_real_from_real): Avoid the operation if
flag_signaling_nans is on and the operand is an sNaN.
(integer_valued_real_unary_p): Update comment stating it
returns false for sNaN values.
(integer_valued_real_binary_p, integer_valued_real_call_p): Same.
(integer_valued_real_single_p): Same.
(integer_valued_real_invalid_p, integer_valued_real_p): Same.
* fold-const-call.c (fold_const_pow): Avoid the operation
if flag_signaling_nans is on and the operand is an sNaN.
(fold_const_builtin_load_exponent) Same.
(fold_const_call_sss): Same for CASE_CFN_POWI.
* gimple-fold.c (gimple_assign_integer_valued_real_p): Same.
(gimple_call_integer_valued_real_p): Same.
(gimple_phi_integer_valued_real_p): Same.
(gimple_stmt_integer_valued_real_p): Same.
* simplify-rtx.c (simplify_const_unary_operation): Avoid the
operation if flag_signaling_nans is on and the operand is an sNaN.
(simplify_const_binary_operation): Same.
* tree-ssa-math-opts.c (gimple_expand_builtin_pow): Avoid the
operation if flag_signaling_nans is on and the operand is an sNaN.
projects / thirdparty / gcc.git / log
